TY - JOUR
T1 - Self-Tuning Transfer Dynamic Convolution Autoencoder for Quality Prediction of Multimode Processes With Shifts
AU - Yang, Chao
AU - Liu, Qiang
AU - Wang, Chen
AU - Ding, Jinliang
AU - Cheung, Yiu ming Cheung
N1 - This work was supported in part by the National Natural Science Foundation of China under Grant U20A20189, Grant 61991401, Grant 62161160338, and Grant U23A20328, in part by the National Natural Science Foundation of China (NSFC)/Research Grants Council (RGC) Joint Research Scheme under Grant N_HKBU214/21, in part by the General Research Fund of RGC under Grant 12201321, Grant 12202622, and Grant 12201323, and in part by the RGC Senior Research Fellow Scheme under Grant SRFS2324-2S02.
Publisher Copyright:
IEEE
PY - 2024/9
Y1 - 2024/9
N2 - Process shift of multimode process involving data distribution and dynamic relation makes traditional transfer learning methods be intractable and even result in negative transfer. To tackle this issue, this article proposes a novel self-tuning transfer dynamic modeling method for quality prediction of multimode processes. First, in order to capture domain-invariant spatiotemporal (DIST) features, a transfer dynamic convolution autoencoder (TDCAE) with a feature decomposition structure is established. Meanwhile, a first-order vector autoregressive constraint is embedded to extract consistent inner dynamics for DIST features. Then, a shared regression network is established to extract the relations with quality variables. Furthermore, by making full use of private spatiotemporal information from target labeled samples in response to the process shift, the self-tuning TDCAE (STDCAE) aided by a fine-tuning strategy is established for online compensation. Finally, the efficacy of the proposed TDCAE and STDCAE is demonstrated by a comprehensive study of a three-phase flow facility process.
AB - Process shift of multimode process involving data distribution and dynamic relation makes traditional transfer learning methods be intractable and even result in negative transfer. To tackle this issue, this article proposes a novel self-tuning transfer dynamic modeling method for quality prediction of multimode processes. First, in order to capture domain-invariant spatiotemporal (DIST) features, a transfer dynamic convolution autoencoder (TDCAE) with a feature decomposition structure is established. Meanwhile, a first-order vector autoregressive constraint is embedded to extract consistent inner dynamics for DIST features. Then, a shared regression network is established to extract the relations with quality variables. Furthermore, by making full use of private spatiotemporal information from target labeled samples in response to the process shift, the self-tuning TDCAE (STDCAE) aided by a fine-tuning strategy is established for online compensation. Finally, the efficacy of the proposed TDCAE and STDCAE is demonstrated by a comprehensive study of a three-phase flow facility process.
KW - Convolutional neural networks
KW - deep autoencoder
KW - dynamic process modeling
KW - multimode processes
KW - quality prediction
KW - transfer learning (TL)
UR - http://www.scopus.com/inward/record.url?scp=85194863117&partnerID=8YFLogxK
U2 - 10.1109/TII.2024.3399932
DO - 10.1109/TII.2024.3399932
M3 - Journal article
AN - SCOPUS:85194863117
SN - 1551-3203
VL - 20
SP - 11295
EP - 11305
JO - IEEE Transactions on Industrial Informatics
JF - IEEE Transactions on Industrial Informatics
IS - 9
ER -